Classification framework for faulty-software using enhanced exploratory whale optimizer-based feature selection...

Area of Science:

• Software Engineering
• Machine Learning
• Data Mining

Background:

• Software Fault Prediction (SFP) is crucial for early detection of defects in software development.
• Existing machine learning techniques require optimized feature selection for effective SFP.

Purpose of the Study:

• To propose a novel machine learning framework for Software Fault Prediction (SFP).
• To enhance feature selection for SFP using metaheuristic optimization algorithms.

Main Methods:

• Pre-processing and re-sampling of SFP datasets.
• Comparison of seven classifiers: KNN, NB, LDA, LR, DT, SVM, and RF.
• Development of the SBEWOA method by hybridizing Binary Whale Optimization Algorithm (BWOA) with Grey Wolf Optimizer (GWO) and Harris Hawks Optimization (HHO) for feature selection.

Main Results:

• The Random Forest (RF) classifier demonstrated superior performance in feature elimination.
• The proposed SBEWOA method significantly outperformed nine established feature selection techniques on PROMISE repository datasets.
• SBEWOA achieved competitively superior results in terms of accuracy, feature count, and fitness function.

Conclusions:

• The SBEWOA method is an efficient and effective machine learning approach for Software Fault Prediction.
• The proposed framework offers a significant advancement in identifying faulty software modules early in the development lifecycle.
• SBEWOA provides a robust alternative for similar software engineering problems requiring optimized feature selection.